Extends ball recovery probability by incorporating spatial pitch-control features from 360 data — average team control near the ball, transition probabilities, and most-relevant-player control areas — to capture how defensive positioning affects likelihood of winning the ball back. The difference between the pitch-control-adjusted prediction and the baseline prediction isolates the positioning contribution to defensive recovery.
Add 360-derived features: Voronoi-based pitch control around the ball, control by nearest defender, control by most relevant pressing player. The "Pre-Recovery PC" model includes these; the baseline "Pre-Recovery" model does not. The difference = positioning contribution.